Accepted Manuscript

Title: Does Attachment Security to a Human Handler

Influence the Behavior of Dogs Who Engage in Animal
Assisted Activities?

Authors: Shelby H. Wanser, Monique A.R. Udell

PII: S0168-1591(18)30561-6
DOI: https://doi.org/10.1016/j.applanim.2018.09.005
Reference: APPLAN 4720

To appear in: APPLAN

Received date: 26-10-2017

Revised date: 25-9-2018
Accepted date: 26-9-2018

Please cite this article as: Wanser SH, Udell MAR, Does Attachment
Security to a Human Handler Influence the Behavior of Dogs Who Engage
in Animal Assisted Activities?, Applied Animal Behaviour Science (2018),
https://doi.org/10.1016/j.applanim.2018.09.005

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Does Attachment Security to a Human Handler Influence the Behavior of Dogs Who Engage
in Animal Assisted Activities?

Shelby H. Wanser, Monique A. R. Udell

Department of Animal and Rangeland Sciences, Oregon State University, 112 Withycombe Hall,
2921 SW Campus Way, Corvallis, OR, 97331, USA

Corresponding author – Shelby H. Wanser, Email: shelby.wanser@oregonstate.edu,

PT
Phone: (541) 737-3431, Fax (541) 737-4174

RI
Highlights

 Overall, dogs with secure or insecure attachment behave similarly in AAA setting.

SC
 Trend suggests AAA dogs with insecure attachment may gaze at handler more.

 Gaze may be a contact maintaining behavior when physical proximity is not possible.

U
 Gazing may provide insight into motivation and coping of dogs in AAA setting.
N
A
ABSTRACT
M

Pet and working dogs raised with humans are known to form attachments to their caregivers and

other humans with whom they have a stable relationship. Attachment style varies across dog-
D

human dyads, with securely attached dogs exhibiting the secure base effect, an ability to find
TE

comfort in the presence of an attachment figure in unusual situations, allowing for greater

exploration. The secure base effect is also known to facilitate interactions with unfamiliar
EP

individuals. Dogs who engage in Animal Assisted Activities (AAA) are often asked to engage with
CC

unfamiliar people in unfamiliar environments, therefore it is possible that dogs with a secure

attachment to their human handlers may be more prepared for success in this role. This study
A

evaluated the behavior of 16 dogs who engage in AAA. Using a secure base test dogs were

categorized as demonstrating secure (exhibiting the secure base effect with their owner/handler; n

= 8) or insecure (not exhibiting the secure base effect; n = 8) attachment styles towards their

handlers. Later the dyads participated in a mock animal assisted activity session to evaluate their
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

working behavior. Our findings indicate that independent of attachment style, dogs who engage in

AAA spent significantly more time in proximity to, and touching, the AAA participant than their

handler (p < 0.001 for both proximity and touch). However, on average the AAA dogs spent

significantly more time gazing at their handler than at the participant during the session (p = 0.03).

Dogs with an insecure attachment style appear to have driven this effect as evidenced by a non-

PT
significant trend suggesting that they gazed longer at their handlers than at the participant (p =

RI
0.07), whereas secure dogs did not display the same trend (p = 0.24). This could suggest that while

their training mandates proximity and interaction with unfamiliar people, dogs who engage in AAA

SC
may be using gaze to maintain contact with their handlers, especially in the absence of a secure

attachment where prolonged comfort seeking from the attachment figure would be expected.

U
N
Keywords: Animal Assisted Activities; Dogs; Secure Base Effect; Attachment; Gaze
A
M

1. Introduction

While it is well established that on average dogs and humans are capable of sharing strong
D

mutual bonds, research exploring individual differences in the dog-human relationship is especially
TE

important in applied contexts (Udell and Brubaker 2016), for example when trying to predict the

success and welfare of individual dogs assigned to a working role. Evaluating dog attachment
EP

styles to owners or handlers provides one method for understanding how relationship patterns could

affect behavioral outcomes in a variety of contexts, especially as it relates to the presence or absence
CC

of the secure base effect. The secure base effect is a byproduct of a secure attachment bond that

provides the “experience of security and comfort obtained from the relationship with the partner,
A

and yet the ability to move off from the secure base provided by the partner, with confidence to

engage in other activities” (Ainsworth 1989, p.711). For individuals with secure attachments, the

presence of the attachment figure enhances the likelihood that the individual will explore and

engage with the environment (Bowlby 1982). The ability to confidently navigate unusual

2
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

situations, meet strangers, or adjust to new environments is often an important consideration for

working dogs, especially dogs who engage in Animal Assisted Activities (AAA) who are tasked

with visiting a wide range of individuals across a variety of settings.

The presence of attachment bonds between dogs and their owners has been well

PT
documented (Topál et al. 1998; Palmer and Custance 2008; Mariti et al. 2013), however only a few

studies to date have categorized dogs into formal attachment styles (Schöberl et al. 2016; Thielke

RI
et al. 2017) or evaluated the influence of attachment security on other behaviors (Horn, Huber and

SC
Range 2013; Thielke et al. 2017). While attachment research has often been associated with the

human developmental literature, especially the Ainsworth Strange Situation Test (SST) originally

U
developed to assess attachment style between infants and mothers, attachment bonds and the secure
N
base effect are behavioral patterns common in nature and have been demonstrated using a variety
A
of methods in many species that require parental care across a wide range of taxa (Gubernick 1981;
M

Kruijt 1984). At least one experimental assessment also predates the Ainsworth SST (Bowlby

1960). In the 1950’s Harry Harlow conducted a test that evaluated the behavior of infant macaques
D

when placed in an unfamiliar room with unfamiliar play objects and either their cloth surrogate
TE

mother (a rough inanimate model of a monkey covered in artificial fur that the infant had previously

been imprinted to) or alone. Harlow found that when the infant was alone in the room, they would
EP

often freeze and huddle in a crouched position for the duration of the test or sometimes they would

run frantically around the room and vocalize in distress. On the other hand, when the cloth mother
CC

was present, the infant would run to her and cling to her for a while, and then would move away to

explore the room or engage with objects in the room, demonstrating a contact-exploration balance.
A

He concluded that the infants were using the surrogate cloth mother as “a source of security, a base

of operations” (Harlow 1958), which became the foundation for future exploration of the secure

base effect in both humans and non-human animals. Recently the secure base test has been adapted

for use with dogs allowing for classification of dogs into attachment styles categories, and posits

3
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

several advantages compared with the SST including shorter testing time (6 versus 21 minutes) and

elimination of order effects associated with the SST (Thielke et al. 2017).

Research has also shown that dogs can provide social support for humans (Cusack 1988;

Crawford, Worsham and Swinehart 2006; Zilcha-Mano, Mikulincer and Shaver 2012). This

PT
strength of dog-human interactions is why dogs are sometimes used to promote the well-being of

people in a variety of settings in the form of Animal Assisted Interventions. Animal Assisted

RI
Activity (AAA) is one common type of animal assisted intervention. According to the International

SC
Association of Human-Animal Interaction Organizations (IAHAIO), Animal Assisted Activity is

defined as “a planned and goal oriented informal interaction and visitation conducted by the human-

U
animal team for motivational, educational and recreational purposes” (IAHAIO 2018). AAA may
N
be conducted in a variety of settings to accomplish a range of purposes, including improving the
A
quality of life of someone in a hospital, assisted-living facility, rehabilitation center, or correctional
M

facility (Arkow 1984; Tsai, Friedmann and Thomas 2010; Burch 2015; IAHAIO 2018), or

“providing comfort and support for trauma, crisis and disaster survivors” (IAHAIO 2018). Dogs
D

who engage in AAA are often pet dogs trained by their owner, whom they work alongside during
TE

AAA.
EP

Conducting AAA with dogs generally involves an owner/handler directing their dog to

engage with an unfamiliar AAA participant. This may involve the dog approaching and touching
CC

the unfamiliar person, allowing them to pet him/her or initiate play. At other times the AAA

participant may sit passively in the presence of the dog or talk to the dog. Training for AAA
A

requires dogs to demonstrate excellent obedience, an aptitude for engaging with unfamiliar people,

and appropriate and reliable responses to potential stimuli including unfamiliar places, objects,

sounds, and smells, and atypical behavior from adults and children, such as unusual touch and

petting behavior and vocalizations (Therapy Dogs International 2015). Given the nature of this

4
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

work, it is critical that dogs who engage in AAA not only feel comfortable in novel or ‘strange’

environments but are comfortable leaving the proximity of their owner/handler in such situations

to interact with unfamiliar individuals.

Research conducted on human infant attachment styles has demonstrated that individuals

PT
with a secure attachment style are less disturbed by the presence of a stranger and are more likely

to interact with them, in contrast with individuals with an insecure attachment style (Bowlby 1982).

RI
Consequently, it seems possible that dogs who engage in AAA who have a secure attachment style

SC
might be at an advantage in their working role when compared to AAA dogs with an insecure

attachment style. To date, no study has investigated whether a dog’s behavior in an AAA setting

U
is impacted by their attachment style, and more specifically, whether or not they exhibit the secure

base effect with their owner/handler. N

A
M

In the present study we asked if and how the secure base effect influenced an AAA dog’s

behavior towards their handler and towards an AAA participant in a mock animal assisted activity
D

setting. Given the training requirements of the working role, we expected that all dogs who engaged
TE

in AAA would spend a substantial amount of time interacting with the unfamiliar AAA participant.

However we predicted that AAA dogs exhibiting the secure base effect with their handler might
EP

spend a greater proportion of session time in proximity to, and gazing at, the AAA participant (as

opposed to their handler) than dogs that did not exhibit the secure base effect with their handler.
CC

2. Methods
A

2.1. Participants

Sixteen dogs who engage in AAA were recruited through personal contact, and

advertisement on a local therapy dog listserv. Dogs comprised ten males (five neutered prior to

study; one neutered between the two sessions of the study) and six females (all spayed), ranging in

5
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

age from six months to twelve years. Participants included three golden retrievers, two German

shepherds, two whippets, one Australian shepherd, one Labrador retriever, one Shiloh shepherd,

one smooth collie, one Pembroke Welsh corgi, one miniature poodle, and three mixed-breeds.

Dogs’ experience engaging in AAA ranged from two months to eight years. In all cases the AAA

dog’s handler was also its owner, with the exception of the 8-year-old mixed breed (Labrador

PT
retriever and wirehaired pointer) who lived with a neighbor, but were highly familiar with one

another and had a well-developed working relationship. From this point forward the owner/handler

RI
will be referred to simply as the handler.

SC
2.2. Ethical statement

U
AAA dog-handler dyads participated on a voluntary basis. Informed consent was obtained
N
from the handlers. No data were collected on any humans involved. The methods associated with
A
this study were approved by the IACUC (ACUP #4444) of Oregon State University. All mandatory
M

laboratory health and safety procedures were complied with in the course of conducting this

experiment. There were no conflicts of interest or competing interests associated with this research.
D
TE

2.3. Testing area

The testing area was a sparsely furnished room measuring 4 m by 4.5 m. The room had
EP

one window and two doors but only one door was used for entrance and exit. The secure base test

methods and testing area layout were based on the canine version of the test conducted by Thielke
CC

et al. (2017). One chair was located along a wall approximately 3 m from the door. A circle with

a 1 m radius was measured around the chair with tape (see Figure 1). A tripod and video camera
A

were located in the corner opposite the door and chair. Three toys – a tennis ball, a squeaky toy,

and a rope toy – were spread out on the floor outside of the taped circle. This room and setup was

novel to the dog and handler prior to testing.

6
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

A mock animal assisted activity session was conducted on a later date (four to thirty days

later) in the same room. For this session, two chairs were located facing each other at a right angle

in the corner of the room opposite the door. Two tangential circles (each with a 0.66 m radius)

were taped on the floor, one around each chair (see Figure 2). The tripod and video camera were

near the door, opposite from the chairs.

PT
2.4. Secure base test

RI
There were two female experimenters: E1 provided instructions at the start of each

SC
condition to ensure consistent handler behavior (E1 remained outside of the room during all

conditions). E2 stood neutrally/inattentively in the corner of the room controlling the video camera

U
(except for during the alone condition during which the camera was set on a tripod facing the door).
N
The secure base test was divided into three conditions, each lasting two minutes.
A
M

2.4.1. Baseline condition

The experimenter led the AAA dog and handler into the room and indicated for the handler
D

to remove the dog’s leash and take a seat in the chair. The handler was instructed that when the
TE

dog entered the circle surrounding their chair, they could interact with the dog (i.e. talking, petting,

playing), but when the dog was outside the circle, they must remain silent, passive, and non-moving
EP

(no talking, no gestures).

CC

2.4.2. Alone condition

E1 opened the door to indicate to the handler to stand up, say “goodbye” to the dog, and
A

exit the room. E2 placed the camera on the tripod and left it filming in the direction of the door

and followed the experimenter and handler out of the room, leaving the dog alone. The primary

purpose of the alone phase was to serve as a mild stressor, which would allow for assessment of

the secure base effect during the return condition.

7
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

2.4.3. Return condition

E1 directed the handler to enter the room and sit back down in the chair. E2 followed

closely behind the handler in entering the room and returned to the corner to control the camera,

without any interaction with the dog. The handler’s instructions were the same as for the baseline

PT
condition: when the dog entered the circle surrounding their chair, they could interact with the dog,

but when the dog was outside the circle, they were to remain silent, passive, and non-moving.

RI
SC
2.5. Mock animal assisted activity session

At a later date, four to thirty days after the secure base test, the dog-handler dyads returned

U
to the same testing location for a mock animal assisted activity session. A mock session was chosen
N
over in situ sessions to ensure experimental control over the session location, context, and duration
A
allowing for consistency across dogs/experimental groups, something that might not be ethically
M

possible within the context of genuine AAA sessions. Two females of college age served as the

mock AAA participants across all dogs, determined by scheduling availability. These individuals
D

were unfamiliar to both the dog and handler at the time of the session.
TE

E1 provided instructions prior to the session. Session protocol was determined based on
EP

handler reports of what their typical AAA sessions looked like, which were then integrated into

one standard methodology that could be used consistently across all dogs. The mock AAA
CC

participant (E3) was seated in one of the two chairs prior to the dog’s arrival. The handler was

instructed to enter the room with their dog, introduce themselves and the dog to E3, give the dog
A

the greeting/release command they would typically use on an AAA visit, and then sit down in the

chair next to E3. The session lasted for seven minutes with the first minute as a greeting phase

(when the AAA dog and handler entered the room and greeted the mock AAA participant;

important because during this minute the handler was moving within the space and initiating

8
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

interactions with both the dog and E3). The remaining six minutes consisted of the interaction

phase of the mock AAA session during which the dog’s behavior was video recorded and later

analyzed. The dog remained on leash (handler holding the end of the leash) for the entire session,

but was to have freedom of movement between and around the handler and E3 to the extent the

leash’s length permitted (about 2 m). This was done to remain consistent with the AAA practices

PT
these dyads would typically encounter in their working roles with actual AAA participants. The

handler and E3 were required to stay seated in their designated chairs located within the taped

RI
circles on the floor for the duration of the session. For the three small dogs (less than twenty-five

SC
pounds), both handler and E3 sat on the floor in the same spots that the chairs were located with

larger dogs. E3 was allowed to casually talk to the dog and with the handler, as would be the case

U
in an actual AAA session, as they engaged in petting and interacting with the dog. The handler
N
was instructed to engage with their dog and the participant (E3) in the session as closely as to how
A
they would in a normal AAA session, while remaining in the designated location. The handler, E1,
M

and E3 were all blind to the attachment categorization of the dog at the time of the mock AAA

session.
D
TE

2.6. Behavior coding

2.6.1. Secure base test

EP

All sessions were videotaped using a Kodak Playtouch Zi10 video camera, manufactured

by Eastman Kodak Company, Rochester, NY, USA. Two coders, with prior training in evaluating
CC

canine attachment styles, independently viewed the video recordings of the return phase and

independently categorized the dog’s behavior according to canine attachment style categories
A

previously described in the literature (Schöberl et al. 2016; Thielke et al. 2017): secure, insecure-

avoidant, insecure-ambivalent, and insecure-disorganized (see Table 1). We found 93.75%

independent inter-rater agreement for attachment style categorization. Categorization

disagreements were then jointly reviewed to come to consensus for the final attachment style

9
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

designation using the standard procedure for holistic canine attachment style categorization

(Thielke et al. 2017). The broader categorization of secure or insecure attachment, indicating the

presence or absence of the Secure Base Effect, was the primary focus in this study.

2.6.2. Mock animal assisted activity session

PT
Video of the dog’s behavior during the six-minute interaction phase of the mock AAA

session was analyzed and coded using JWatcher Version 1.0 coding software, developed at

RI
Macquarie University, Sydney, Australia. Each video was coded twice: once for the interaction

SC
between the dog and the mock AAA participant and a second time for the interaction between the

dog and the handler. The behavioral states recorded were: gazing at the participant, touching the

U
participant, participant proximity-seeking (entering the circle taped around the participant’s chair),
N
gazing at the handler, touching the handler, and handler proximity-seeking (entering the circle taped
A
around the handler’s chair). The proportion of time spent engaging in each behavior during the
M

six-minute period was recorded. Proximity and touching were not treated as mutually exclusive.
D

Videos were analyzed by two independent coders. Inter-observer reliability on proportion

TE

of total time for each behavior was calculated using Pearson Correlation Coefficients. There was

strong agreement for all behavioral measures (gaze at handler, R = 0.85; gaze at participant, R =
EP

0.75; touching handler, R = 0.83; touching participant, R = 0.99; proximity to handler, R = 0.89;

proximity to participant, R = 0.99). All data used in the analysis were determined by Coder 1.
CC

2.7. Statistical analysis

A

A Kolmogorov-Smirnov test demonstrated an absence of a normal distribution in the

behavior of all the dogs as a whole in terms of their touch, proximity, and gaze toward the handler

versus the AAA participant, therefore Wilcoxon Signed-Rank tests were used to compare these

measures. Unpaired t-tests were used to assess any differences in age or AAA experience between

10
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

the dogs categorized as having a secure versus insecure attachment style. Unpaired t-tests were

also used to assess any differences between the secure versus insecure dogs in terms of their time

spent touching their handler, touching the participant, in proximity to their handler, in proximity to

the participant, gazing toward their handler, and gazing toward the participant. Additionally, paired

t-tests were used to assess within-group differences in the dogs’ behavior toward the handler versus

PT
the participant (secure dogs – touch, insecure dogs – touch, secure dogs – proximity, insecure dogs

– proximity, secure dogs – gaze, insecure dogs – gaze). The alpha level was set to p < 0.05 for all

RI
tests.

SC
3. Results

U
3.1. Secure base test
N
Based on the canine attachment style definitions presented in Table 1, eight dogs were
A
categorized as displaying a secure attachment and eight dogs were categorized as having an
M

insecure attachment to their handler. Seven dogs in the insecure category demonstrated an

insecure-ambivalent attachment style and one demonstrated an insecure-disorganized attachment

D

style.
TE

After categorization, the age and work experiences of the two groups – secure and insecure
EP

– was compared to ensure these factors were roughly equivalent and could not account for possible

differences in the AAA setting. The mean age of dogs in the secure group was 5.6 years (SD =
CC

3.0) and in the insecure group was 6.9 years (SD = 4.5). There was no significant difference in the

ages of the dogs in the secure versus insecure groups (unpaired t-test t (14) = 0.65, p = 0.53). The
A

mean number of years of AAA experience for dogs in the secure group was 2.8 years (SD = 2.6)

and insecure group was 2.3 years (SD = 2.3). There was no significant difference in the AAA

experience of the dogs in the secure versus insecure groups (unpaired t-test t (14) = 0.36, p = 0.73).

11
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

3.2. Mock animal assisted activity session

First the behavior of all AAA dogs as a group was analyzed to determine if different

patterns of behavior towards their handler versus an AAA participant would be observed in the

AAA setting, since we predicted that given the training associated with this working role, dogs who

engage in AAA should spend more time with the AAA participant than with the handler

PT
independent of attachment style. A Kolmogorov-Smirnov test did not confirm a normal

distribution so a Wilcoxon Signed-Rank test was used. As predicted, dogs who engage in AAA

RI
spent a significantly greater proportion of time in physical contact with the AAA participant than

SC
their handler (median proportion of time in physical contact with participant = 0.86, handler = 0.08;

Z = -3.52, p < 0.001) and in proximity to the AAA participant compared to their handler (median

U
proportion of time in proximity to participant = 0.85, handler = 0.17; Z = -3.36, p < 0.001).
N
However, dogs who engage in AAA spent a significantly greater proportion of session time gazing
A
at their handler compared to session time spent gazing at the AAA participant (median proportion
M

of time spent gazing at participant = 0.09, handler = 0.14; Z = -2.12, p = 0.03).

D

Next the in-session behavior of AAA dogs categorized as secure or insecure was compared.
TE

Both secure and insecure dogs spent significantly more time in physical contact with the AAA

participant than with their handler (secure dogs, paired t-test t (7) = 14.34, p < 0.0001; insecure
EP

dogs, paired t-test t (7) = 7.52, p = 0.0001). There was not a statistically significant difference

between insecure and secure dogs in terms of the proportion of time spent touching the handler
CC

(unpaired t-test t (14) = 0.91, p = 0.38) or the participant (unpaired t-test t (14) = 0.89, p = 0.39),

see Figure 3.
A

Dogs in both the secure and insecure categories spent significantly more time in proximity

to the participant than the handler (secure dogs, paired t-test t (7) = 11.77, p < 0.0001; insecure

dogs, paired t-test t (7) = 3.04, p = 0.02). There was not a statistically significant difference between

12
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

the groups in terms of the proportion of time spent in proximity to the handler (unpaired t-test t (14)

= 0.70, p = 0.49) or the participant (unpaired t-test t (14) = 0.75, p = 0.47), see Figure 4.

In terms of gaze, however, there was a non-significant trend suggesting that dogs in the

insecure group spent more time gazing at their handler than did the dogs in the secure group

PT
(unpaired t-test t (14) = 1.90, p = 0.08). No such trend between the insecure and secure groups was

found for the dogs’ gazes towards the AAA participant (unpaired t-test t (14) = 0.33, p = 0.75).

RI
Additionally, there was a non-significant trend suggesting that the insecure dogs spent more time

SC
gazing at their handler than at the AAA participant (paired t-test t (7) = 2.12, p = 0.07), whereas

there was no such trend in the proportion of time the secure dogs spent gazing at the handler versus

U
at the participant (paired t-test t (7) = 1.30, p = 0.24), see Figure 5. The dog’s lifetime experience
N
in animal assisted activities had no impact on time spent gazing at the handler during the session
A
(pearson correlation r = -0.2053, n = 16, p = 0.45).
M

4. Discussion
D

Overall the patterns of behavior demonstrated by the securely and insecurely attached AAA
TE

dogs were more similar than different within the mock-AAA setting. The results of this study

suggest that, in general, AAA dogs spend a greater proportion of time in proximity to, and in
EP

physical contact with, the AAA participant than their handler during an AAA session, independent

of attachment style towards their handler. Therefore within the current population, dogs that did
CC

not exhibit the secure base effect with their handler appeared to be equally effective in carrying out

the basic requirements of their working role as dogs that exhibited the secure base effect. However,
A

evaluating the gaze of the AAA dogs illuminated an interesting trend that, although not quite

statistically significant, suggests that insecurely attached dogs (dogs who do not exhibit the secure

base effect) may spend more time than securely attached dogs gazing at their handler during an

AAA session.

13
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

Dogs who engage in AAA have been specifically trained to interact with unfamiliar AAA

participants by remaining close to the stranger and allowing that person to pet them. Thus, it is not

surprising that AAA dogs demonstrated greater time spent in proximity to, and touching, the AAA

participant than the handler; their training history may outweigh the influence of attachment style

PT
towards their handler in this case. However, an AAA dog’s gaze to one person or the other during

an AAA session is not typically instructed by the handler, therefore it is possible that gaze could

RI
be a more honest behavioral signal in this context. Consequently it may be worth considering if

SC
“gaze” should be used as another standard measure of contact or proximity-seeking behavior when

evaluating behavioral cues of attachment in applied contexts; especially in cases where physical

U
proximity is incompatible with a trained working role (as is the case for AAA dogs) or is physically
N
impossible (e.g. search and rescue dogs working at a distance).
A
M

While further interpretations of a dog’s gaze need to be made with care, it is possible that

the secure base effect is a determining factor in directional gazes during AAA sessions.
D

Interestingly, in Mary Ainsworth’s research on attachment styles in children she observed that
TE

children often kept their eyes and posture oriented towards their mother while being held by an

unfamiliar person (Bowlby 1982). Gazing may be a form of social referencing, the seeking of
EP

information from another individual from which to base one’s response to a stimulus or from which

to base one’s actions. For example the dog may be gazing at their handler to gain information about
CC

how to interact with the other person or other stimuli in the room. In one study, Merola, Prato-

Previde and Marshall-Pescini (2012) found that while dogs alternated their gaze between their
A

owner and a stranger similarly, dogs based their behavior more on cues received from their owner

than from the stranger. Consistent with the secure base effect, the handler’s presence would give

the secure dogs the base from which to perform the AAA work. The secure dogs may not need to

refer to their handler for much reinforcement or guidance, so they could be expected to gaze at each

14
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

person for roughly similar amounts of time as Merola, Prato-Previde and Marshall-Pescini (2012)

demonstrated. In contrast, the insecure dogs may gaze at the handler more often due to a greater

need for reassurance or behavioral guidance.

Another recent study concluded that both dogs and owners experience increased urinary

PT
oxytocin concentrations as a result of mutual gazing. The duration of time the dog spent gazing at

the owner was proportional to the change in the oxytocin level of both the dog and owner

RI
(Nagasawa et al. 2015). Given that all of the AAA dogs spent some time gazing at their

SC
owner/handler, we could expect that this behavior served to increase oxytocin levels in both

themselves and their owner/handler, which may serve to reinforce the gazing behavior or possibly

U
the AAA work as a whole. This may also represent another mechanism by which gazing at the
N
owner/handler could reduce stress that dogs, especially insecure dogs, may experience when
A
engaging with an unfamiliar person in an AAA setting.
M

5. Limitations and future directions

D

The current study provides the first behavioral evidence that the attachment style of an
TE

AAA dog to their handler may influence some aspects of their human-directed interactions in a

working context. Future studies are needed to determine if the trends identified here are present in
EP

larger populations and across different types of animal assisted interventions, or if attachment style

holds more predictive value for dogs engaged in some activities versus others.
CC

Considering that individuals with an insecure attachment style typically prefer to seek the
A

proximity of bonded individuals in unfamiliar settings, prolonged interaction with an unfamiliar

AAA participant by the insecure dogs is quite likely the result of training instead of an independent

drive to engage with the unfamiliar person. If so, another potential consideration for future research

is whether animal assisted activities result in higher stress levels for some animals (i.e. those with

15
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

an insecure attachment style) more than others, even if both show similar levels of interaction with

the AAA participant. This could have important welfare implications and should be further

evaluated using additional behavioral and physiological measures of stress (e.g. cortisol levels

during AAA sessions). Data collected during actual AAA sessions, as opposed to mock sessions,

would also be valuable although likely would require much larger sample sizes due to more variable

PT
interactions in naturalistic settings.

RI
6. Conclusions

SC
There is a great need for more research evaluating the behavior and welfare of the animals

participating in working roles (Cobb et al. 2015), including animal assisted activities. Such

U
research is important for a number of reasons. First, animals are most likely to perform their jobs
N
well if they are well suited to the task and do not show signs of stress within their working
A
environment. Second, the comfort level of an animal in many working roles may have implications
M

for the safety and wellbeing of the humans involved. This may be especially true in AAA settings

where animals are expected to work with vulnerable populations, where injury or even rejection by
D

a nervous animal could have serious implications. However there are equally compelling reasons
TE

to consider the behavior and the wellbeing of working animals in their own right. Doing so will

allow us to better understand what aspects of their lives they are well suited for, which aspects
EP

might be causing stress or even injury, as well as what knowledge, methods, and practices result in

positive welfare outcomes for both dogs and people, as well as the best human-animal interactions
CC

possible.

We wish to confirm that there are no known conflicts of interest associated with this publication
A

and there has been no significant financial support for this work that could have influenced its
outcome.

Acknowledgements

16
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

We are grateful to the dogs and their handlers who volunteered for this study. We would

like to thank Rachel Webber, Megan Hughes, and Erika Nivens for their assistance with data

collection and video coding.

PT
RI
SC
U
N
A
M
D
TE
EP
CC
A

17
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

References

Ainsworth, M. D. S. 1989. Attachments beyond infancy. American Psychologist 44: 709-716.

Arkow, P. 1984. Dynamic Relationships in Practice: Animals in the Helping Professions.

Alameda, California: Latham Foundation.

Bowlby, J. 1960. Separation Anxiety. International Journal of Psycho-Analysis. 41: 89-113.

PT
Bowlby, J. 1982. Attachment and Loss, Vol. 1: Attachment. 2nd edn. New York: Basic Books.

Burch, M. 2015. Does your dog have what it takes to be a therapy dog? Getting started in
Animal-Assisted Therapy. Raleigh, North Carolina: American Kennel Club.

RI
Crawford, E. K., Worsham, N. L., and Swinehart, E. R. 2006. Benefits derived from companion
animals, and the use of the term “attachment.” Anthrozoös 19: 98-112.

SC
Cobb, M., Branson, N., McGreevy, P., Lill, A., and Bennett, P. 2015. The advent of canine
performance science: Offering a sustainable future for working dogs. Behavioural

U
Processes 110: 96–104.

N
Cusack, O. 1988. Pets and Mental Health. New York: Routledge.

Gubernick, D. J. 1981. Parent and Infant Attachment in Mammals. In D. J. Gubernick & P. H.

A
Klopfer (Eds.), Parental Care in Mammals (pp. 243–305). Boston, MA: Springer US.
https://doi.org/10.1007/978-1-4613-3150-6_7
M

Harlow, H. F. 1958. The Nature of Love. American Psychologist 13: 673-685.

D

Horn, L., Huber, L., and Range, F. 2013. The Importance of the Secure Base Effect for Domestic
Dogs – Evidence from a Manipulative Problem-Solving Task. PloS One 8: e65296.
TE

International Association of Human-Animal Interaction Organizations. 2018. IAHAIO White

Paper: The IAHAIO definitions for animal assisted intervention and guidelines for wellness
of animals involved in AAI, Retrieved from http://iahaio.org/wp/wp-
EP

content/uploads/2018/04/iahaio_wp_updated-2018-final.pdf

Kruijt, J. P. 1984. On the Development of Social Attachments in Birds. Netherlands Journal of

Zoology 35: 45–62. https://doi.org/10.1163/002829685X00064
CC

Mariti, C., Ricci, E., Zilocchi, M., and Gazzano, A. 2013. Owners as a secure base for their dogs.
Behaviour 150: 1275–1294.
A

Merola, I., Prato-Previde, E., and Marshall-Pescini, S. 2012. Dogs’ social referencing towards
owners and strangers. PloS One 7: e47653.

Nagasawa, M., Mitsui, S., En, S., Ohtani, N., Ohta, M., Sakuma, Y., Onaka, T., Mogi, K., and
Kikusui, T. 2015. Oxytocin-gaze positive loop and the coevolution of human-dog bonds.
Science 348: 333–336.

Palmer, R., and Custance, D. 2008. A counterbalanced version of Ainsworth’s Strange Situation

18
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

Procedure reveals secure-base effects in dog–human relationships. Applied Animal

Behaviour Science 109: 306–319.

Schöberl, I., Beetz, A., Solomon, J., Wedl, M., Gee, N., and Kotrschal, K. 2016. Social factors
influencing cortisol modulation in dogs during a strange situation procedure. Veterinary
Behavior 11: 77-85.

Therapy Dogs International (TDI). Therapy Dogs International Testing Requirements.

www.tdi-dog.org. Accessed on May 7, 2015.

PT
Thielke, L. E., Rosenlicht, G., Saturn, S. R., and Udell, M. A. R. 2017. Nasally-Administered
Oxytocin Has Limited Effects on Owner-Directed Attachment Behavior in Pet Dogs
(Canis lupus familiaris). Frontiers in Psychology 8: 1699.

RI
Topál, J., Miklósi, Á., Csányi, V., and Dóka, A. 1998. Attachment behavior in dogs (Canis
familiaris): A new application of Ainsworth’s (1969) Strange Situation Test.

SC
Comparative Psychology 112: 219–229.

Tsai, C., Friedmann, E., and Thomas, S. 2010. The effect of Animal-Assisted Therapy on stress

U
responses in hospitalized children. Anthrozoös 23: 245-258.

N
Udell, M. A. R. and Brubaker, L. 2016. Are Dogs Social Generalists? Canine Social Cognition,
Attachment, and the Dog-Human Bond. Current Directions in Psychological Science 25:
327-333.
A
Zilcha-Mano, S., Mikulincer, M., and Shaver, P. R. 2012. Pets as safe havens and secure bases:
M

The moderating role of pet attachment orientations. Research in Personality 46: 571–580.
D
TE
EP
CC
A

19
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

Figure captions

Figure 1. Secure base test

Diagram of room set-up for secure base test.

Figure 2. Mock animal assisted activity session

PT
2a. Diagram of room set-up for mock animal assisted activity session. 2b. Image of mock animal

assisted activity session.

RI
SC
Figure 3. Physical contact

Proportion of time dogs spent in physical contact with the owner/handler and the AAA participant

U
during the mock animal assisted activity session, grouped by attachment style. Total session time
N
was six minutes. The box outlines the second and third quartiles. The bold horizontal line in the
A
box interior indicates the median. The whiskers mark maximum and minimum values.
M

Figure 4. Proximity
D

Proportion of time dogs spent in proximity to the owner/handler and the AAA participant during
TE

the mock animal assisted activity session, grouped by attachment style. Total session time was

six minutes. The box outlines the second and third quartiles. The bold horizontal line in the box
EP

interior indicates the median. The whiskers mark maximum and minimum values.
CC

Figure 5. Gaze

Proportion of time dogs spent gazing toward the owner/handler and the AAA participant during
A

the mock animal assisted activity session, grouped by attachment style. Total session time was

six minutes. The box outlines the second and third quartiles. The bold horizontal line in the box

interior indicates the median. The whiskers mark maximum and minimum values.

20
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

Fig 1

PT
RI
SC
U
N
A
M

Fig 2
D
TE
EP
CC
A

21
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

PT
RI
SC
U
N
A
M
D
TE
EP
CC
A

22
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

PT
RI
SC
Fig 3

U
N
A
M
D
TE
EP
CC
A

Fig 4

23
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

PT
RI
SC
U
N
A
M

Fig 5
D
TE
EP
CC
A

24
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

PT
RI
SC
U
N
A
M
D
TE
EP
CC
A

25
RUNNING HEAD: ATTACHMENT SECURITY IN AAA DOGS

Table 1. Attachment Style Definitions (adapted from Schöberl et al. 2016 and Thielke et
al. 2017).

Attachment Style Definition

Dog’s greeting behavior is active, open, and positive. Little or

no resistance to contact or interaction with the owner/handler.
Secure

PT
Seeks proximity and is comforted upon reunion, returning to
exploration or play.

RI
Dog shows little or no visible response to the owner/handler’s
return. Ignores or turns away from the owner/handler but may
Insecure-Avoidant not resist interaction altogether (e.g. laying, sitting, or standing

SC
without physical contact with, out of reach of, or at a distance
from the owner/handler).

Dog shows exaggerated proximity-seeking and clinging

U
behavior (but may struggle if held by owner/handler). Exhibits
a mix of persistent distress with efforts to maintain physical
Insecure-Ambivalent N
contact with the owner/handler and/or physically intrusive
behavior toward the owner/handler. (Dogs who the judges
A
agreed seemed essentially secure but with ambivalent
tendencies were categorized as secure.)
M

Dog exhibits evidence of a strong approach-avoidance conflict

or fear upon reunion (e.g. circling owner/handler, hiding from
sight, rapidly dashing away upon reunion, or “aimless”
D

wandering around the room). A lack of coherent strategy is

Insecure-
TE

shown by contradictory behavior. Dog may show stereotypies

Disorganized
upon reunion (e.g. freezing or compulsive grooming).
“Dissociation” may be observed, that is, still or frozen posture,
staring into space without apparent cause, for at least 20
EP

seconds (in a non-resting, non-sleeping dog).

Judges were unable to reach consensus on the attachment style

CC

Unclassifiable* categorization of the dog. Unclassifiable dogs were excluded

from further analysis on dog attachment.
A

*No dogs in the present study were unclassifiable.

26